THE STUDY SAMPLE CONSISTED OF 790 MALE FELONS PLACED ON PROBATION IN THE DETROIT (MI) RECORDER'S COURT BETWEEN JULY 1969 AND JANUARY 1973. A STEPWISE MULTIPLE REGRESSION ANALYSIS WAS UTILIZED TO PREDICT FELONY RECIDIVISM WITHIN A ONE YEAR PERIOD FROM THE BEGINNING OF PROBATION. HAVING ESTABLISHED AN ESTIMATED RECIDIVISM RATE OF 17.0 PERCENT AND AN OVERALL VIOLATION RATE OF 40.0 PERCENT, ATTENTION WAS THEN TURNED TO DETERMINING WHICH PROBATIONERS WOULD BE CLASSIFIED AS FAILURES AND HOW THEY WOULD FAIL. A FIRST DISCRIMINANT FUNCTION ANALYSIS CLASSIFIED PROBATION OUTCOME INTO THREE CATEGORIES: NO VIOLATION, TECHNICAL VIOLATION, AND RECIDIVISM. A SECOND CLASSIFIED PROBATION OUTCOME INTO EIGHT CATEGORIES INCLUDING 'NO RECIDIVISM' AND SEVEN DIFFERENT TYPES OF RECIDIVISTIC BEHAVIOR. THE REGRESSION ANALYSIS REVEALED THAT THE INDICATORS OF PRIOR CRIMINAL BEHAVIOR WERE THE BEST PREDICTORS OF RECIDIVISM, WITH TIME BETWEEN ARREST AND SENTENCE AND CAPIAS ISSUED PRIOR TO SENTENCE BEING THE STRONGEST. AGE, RACE, SEX, AND JUVENILE RECORD WERE FOUND TO HAVE NEGLIGIBLE IMPACT ON FELONY RECIDIVISM. SOME 78.9 PERCENT OF THE CASES IN THE ORIGINAL SAMPLE AND 85.0 PERCENT OF THE CASES IN THE VALIDATION SAMPLE WERE CORRECTLY CLASSIFIED. THE TWO DISCRIMINANT ANALYSES SHOWED CLEARLY THAT NON-COMPLIANCE WITH THE RULES OF PROBATION (TECHNICAL VIOLATION) IS DISTINCT FROM RECIDIVISM IN THAT BOTH BEHAVIORS CAN BE PREDICTED SEPARATELY FROM THE BACKGROUND VARIABLES, AND THAT DIFFERENT TYPES OF OFFENDERS RECIDIVATE IN DIFFERENT WAYS. IT IS CONCLUDED THAT RECIDIVISM RATES CAN BE PREDICTED WITH A VERY HIGH DEGREE OF ACCURACY AND THAT RATES FOR PROBATIONERS COULD BE SUBSTANTIALLY REDUCED BY ESTABLISHING AND IMPLEMENTING A SENTENCING POLICY BASED ON THE ACQUISITION AND UTILIZATION OF VALID INFORMATION. ALSO CONCLUDED WAS THAT RECIDIVISM IS EXPLAINED PRIMARILY BY AN ORIENTATION TOWARD CRIMINAL BEHAVIOR IN THE INDIVIDUAL AND ONLY SECONDARILY, IF AT ALL, BY EXTERNAL CIRCUMSTANCES IN THE LIFE OF THE CRIMINAL OFFENDER. A BIBLIOGRAPHY IS INCLUDED....ELW